**2. Data and methodology**

*Banking and Finance*

flows of the firm will cover the cost of the investment [6, 7]. This situation will indirectly affect other decisions of the firms such as employment, credit usage, debt repayments, social insurance payments, and other factors. In addition to all these factors, it is inevitable that firms' balance sheet structure and ratios are also affected by uncertainty in an economy. In case of elevated uncertainty, since consumption is postponed, firms' net sales decline, and this negatively influences their profitability ratios. Furthermore, firms' indebtedness might decline because their investment is also postponed and they no longer need to borrow to finance their investment. Firms' cash holdings and liquidity ratios might improve, as they want to stay highly liquid against any negative shocks under rising uncertainty. Last but not least, firms' cost of finance and interest expenses may be negatively affected, as heightening uncertainty might lead to an increase in risk premium and depreciation in the currency where the firms' borrowing is denominated in. All these channels, and possibly more than these, explain the transmission between economic uncertainty and firm balance sheet performance. Motivating from these verities, the main objective of this chapter is to empirically analyze the impact of economic uncertainty on firms' balance sheet performance using the financial statements of real sector firms quoted in Borsa İstanbul. There is a wide range of literature on the relationship between firm balance sheet

performance and overall macroeconomic dynamics for different countries. For instance [8] analyzes the implications of macroeconomic instabilities and institutional factors on the financial distress of Chinese-listed companies. [9] assesses the influence of macroeconomic conditions on the debt currency composition of firms in Brazil and how exchange rate movements impact the balance sheets and investment decisions of firms. In a cross-country setting, [10] questions whether the use of macroeconomic and industrial indicators improves the performance measures of solvent and insolvent firms. Finally, [11] deals with the balance sheet impact of foreign currency debt and exchange rate depreciation on the investments of firms in Korea. In addition, there is a huge literature dealing with the impact of uncertainty on overall economy as well as other dimensions of the economy. One strand of the literature investigates directly the possible influence of uncertainty shocks on economic activity and finds out that uncertainty negatively impacts the aggregate demand [12–16]. Another strand of the literature looks at the issue by empirically analyzing how uncertainty impacts the households' consumption [17–19]. There are also many studies analyzing the implications of uncertainty with firms' perspectives. Although the majority looks at the effect of uncertainty on firms' investment decisions [20–24], some studies focus on the impact of uncertainty on firms' balance sheets. Nguyen, Kim, and Papanastassiou [25] investigate the link between economic policy uncertainty and financial derivative usage of firms in East Asia and find that as uncertainty accelerates, firms use derivative instruments extensively to hedge their risks. Hankins et al. [26] document the relationship between political uncertainty and corporate cash holdings of the US firms and finds out that following an uncertainty shock, cash accounts increase and capital spending declines. A similar study by Feng, Lo, and Chan [27] on Chinese firms claims that firms with higher firm value increase their cash holdings more than other firms as economic policy uncertainty heightens. In their cross-country study with firm-level data set, Gungoraydınoğlu, Çolak, and Öztekin [28] report that firm leverage drops in the wake of rising political uncertainty. In their study on US public corporates, Tran and Phan [29] obtain that policy uncertainty is positively related to the cost of debt and negatively related to maturity of debt. Francis, Hasan, and Zhu [30] find that rising political uncertainty increases the borrowing costs of nonfinancial firms. Iqbal, Gan, and Nadeem [31] document that economic policy uncertainty adversely affects firms' performance proxied by return on equity, return on assets, net profit

**166**

margin, and Tobin's Q.

Our sample consists of the financial statements of 417 nonfinancial firms quoted in Borsa İstanbul between 2005Q1 and 2019Q1 at a quarterly frequency. This corresponds to, on average, 300 firm observations each quarter and 13,356 unique observations in total. Real sector firms are obtained by eliminating the financial sector firms, sport companies, and real estate investment funds from the population of BIST firms. Quarterly income statement items are annualized by aggregating the latest four quarters.

Initially, we measure balance sheet strength of the listed real sector firms in Borsa İstanbul (BIST) with a composite index called Multivariate Firm Assessment Score (MFA score) which combines different corporate finance ratios such as liquidity, leverage, and profitability. This index which have a 90 percent predictive power improves Altman Z-score [32] methodology for Turkish firms. Specifically, using the multivariate discriminant analysis (MDA) methodology for the seven financial ratios explained in **Table 1**, we obtain MFA score for each firm in the sample.

Later on, we removed the firms with MFA scores above the 95th percentile and below the 5th percentile in the entire sample, in order to eliminate the outlier observations. Finally, we take the average of MFA scores and obtain mean MFA scores for each quarter. The descriptive statistics of the ratios in the MFA score and MFA score itself is provided in **Table 2**.

In the analysis, we consider four uncertainty indices: economic uncertainty, financial uncertainty, and consumer uncertainty indices which are developed by Sahinoz and Cosar [5], and trade uncertainty index for the Turkish economy from Ahir, Bloom and Furceri [41]. Consumer uncertainty index is a survey indicator which shows the uncertainty perception of consumers in the economy, while

*MFA score =* **0.24** *X***<sup>1</sup>** *−* **0.14** *X***<sup>2</sup>** *−* **0.03** *X***<sup>3</sup>** *+* **3.76** *X***<sup>4</sup>** *−* **0.72** *X***<sup>5</sup>** *+* **0.20** *X***<sup>6</sup>** *+* **1.14** *X***<sup>7</sup>**

*X* 1 = (*CashEquivalents* + *Securities* + *ShortTermTradeRecievables*)∕(*ShortTermLiabilities*):

This indicator, also known as the acid-test ratio, shows how much the short-term debt of the firm can be met with cash and cash equivalents

*X*<sup>2</sup> = *ShortTermLiabilites*/*CurrentAssets* :

It measures the firm's ability to pay its short-term liabilities with short-term assets

*X*<sup>3</sup> = *TotalLiabilities*/*Equities*

It shows how much sufficient the firm's equities to pay its debt

*X* 4 = *EBITDA*/*TotalAssets*

It is the profitability of the firm from its main activities by asset size

*X*<sup>5</sup> = *FinancialExpenses*/*NetSales*

Indicates the capacity of the company to pay the FX and interest expenses arising from its debts

*X*<sup>6</sup> = *NetProfit*(*Loss*)/*NetSales*

It is the net earnings (or loss) of the firm per sale at the end of the period

*X*<sup>7</sup> = *Retained Earnings*/*Total Assets*

It is the measure of cumulative profit or loss from the past periods. It also contains information about the age of the company

#### **Table 1.**

*MFA score equation and variable definitions.*


#### **Table 2.**

*Descriptive statistics of MFA score variables.*

financial uncertainty index mostly indicates market volatility through global financial conditions. On the other hand, economic uncertainty is a weighted index of consumers, producers, forecasters, and economic policy uncertainty indices in a dynamic factor model framework. Finally, trade uncertainty index measures uncertainty related to trade in the Economist Intelligence Unit country reports. The descriptive statistics of uncertainty indices are exhibited in **Table 3**.

The movement in the economic uncertainty index for Turkish economy over time is depicted in **Figure 1**. It is evident from the figure that there are two specific episodes, where uncertainty has risen sharply. The first one is the global financial crisis from 2008 till the last quarter of 2009 and the other one corresponds to the period after the 2018 foreign exchange market turbulence in Turkey. In this study, we will give particular attention to these episodes to investigate the causality from uncertainty to firm balance sheets.

**169**

parameter VAR:

variables in the model, and *ut*

*The Relationship between Economic Uncertainty and Firms' Balance Sheet Strength*

**Uncertainty indices Obs. Mean Std. Dev. Min Max** Economic 57 0.07 0.83 −0.90 2.60 Financial 57 −0.05 0.93 −1.30 3.20 Consumer 57 0.02 1.13 −1.10 5.90 Trade 57 0.33 0.20 0.00 0.89

Our econometric analysis basically use time-varying Granger causality between an uncertainty index and MFA score based on Rossi and Wang [42] paper. This methodology is more powerful if the series have instabilities. In addition, classical Granger causality test does not allow to drive time-varying parameters. VAR-based reducedform Granger causality test requires stationarity in the data, and its test statistics is not valid if the series have potential structural breaks. Because of the nonstationary nature and existence of structural breaks in **Figure 1**, Rossi and Wang's [42] is the best methodology to analyze time-varying Granger causality with this data.

Following Rossi and Wang [42], Eq. (2) shows a reduced-form time-varying

(*L*) *yt* = *ut* (1)

*iid*(0,Σ) (3)

(*L*) = *I* − *A*1,*tL* − *A*2,*<sup>t</sup> L*<sup>2</sup> − …− *Am*,*<sup>t</sup> L<sup>m</sup>* (2)

*At*

*ut*<sup>~</sup>

 shows error term. The iteration of Eq. (2) and projection of *yt* onto the linear space created by

where *Aj*,*t*,*j* = 1,…*m* show time-varying coefficients, [*yt*, *yt*−1,…, *yt*<sup>−</sup>*m*]′ show

*At*

*Economic uncertainty index for Turkey. Source: Cosar and Sahinoz [5].*

(*y t*−1, *y t*−2,…, *y t*−*m*)′ derive the following equation:

*DOI: http://dx.doi.org/10.5772/intechopen.91860*

*Descriptive statistics of uncertainty indices.*

**Table 3.**

**Figure 1.**


*The Relationship between Economic Uncertainty and Firms' Balance Sheet Strength DOI: http://dx.doi.org/10.5772/intechopen.91860*

**Table 3.**

*Banking and Finance*

with cash and cash equivalents *X*<sup>2</sup> = *ShortTermLiabilites*/*CurrentAssets* :

*X*<sup>3</sup> = *TotalLiabilities*/*Equities*

*X* 4 = *EBITDA*/*TotalAssets*

*X*<sup>5</sup> = *FinancialExpenses*/*NetSales*

*X*<sup>6</sup> = *NetProfit*(*Loss*)/*NetSales*

of the company

**Table 1.**

*X*<sup>7</sup> = *Retained Earnings*/*Total Assets*

*MFA score equation and variable definitions.*

*MFA score =* **0.24** *X***<sup>1</sup>** *−* **0.14** *X***<sup>2</sup>** *−* **0.03** *X***<sup>3</sup>** *+* **3.76** *X***<sup>4</sup>** *−* **0.72** *X***<sup>5</sup>** *+* **0.20** *X***<sup>6</sup>** *+* **1.14** *X***<sup>7</sup>**

*X* 1 = (*CashEquivalents* + *Securities* + *ShortTermTradeRecievables*)∕(*ShortTermLiabilities*):

It measures the firm's ability to pay its short-term liabilities with short-term assets

Indicates the capacity of the company to pay the FX and interest expenses arising from its debts

It shows how much sufficient the firm's equities to pay its debt

It is the profitability of the firm from its main activities by asset size

It is the net earnings (or loss) of the firm per sale at the end of the period

This indicator, also known as the acid-test ratio, shows how much the short-term debt of the firm can be met

**168**

**Table 2.**

uncertainty to firm balance sheets.

*Descriptive statistics of MFA score variables.*

financial uncertainty index mostly indicates market volatility through global financial conditions. On the other hand, economic uncertainty is a weighted index of consumers, producers, forecasters, and economic policy uncertainty indices in a dynamic factor model framework. Finally, trade uncertainty index measures uncertainty related to trade in the Economist Intelligence Unit country reports. The

**Variable Obs. Mean Std. dev. Min Max** Acid-test ratio 13,356 1.43 1.59 0.00 28.68 ST liabilities/current assets 13,356 0.83 0.82 0.03 19.99 Total liabilities/equity 13,356 2.36 6.27 −1.52 134.45 EBITDA/assets 13,356 0.08 0.10 −0.80 0.86 Financial exp./sales 13,356 0.05 0.33 −14.84 6.46 Net profit/sales 13,356 0.02 0.93 −47.80 18.28 Retained earnings/assets 13,356 −0.01 0.43 −5.05 1.23 MFA score 13,356 0.42 1.02 −3.48 3.28

It is the measure of cumulative profit or loss from the past periods. It also contains information about the age

The movement in the economic uncertainty index for Turkish economy over time is depicted in **Figure 1**. It is evident from the figure that there are two specific episodes, where uncertainty has risen sharply. The first one is the global financial crisis from 2008 till the last quarter of 2009 and the other one corresponds to the period after the 2018 foreign exchange market turbulence in Turkey. In this study, we will give particular attention to these episodes to investigate the causality from

descriptive statistics of uncertainty indices are exhibited in **Table 3**.

*Descriptive statistics of uncertainty indices.*

#### **Figure 1.**

*Economic uncertainty index for Turkey. Source: Cosar and Sahinoz [5].*

Our econometric analysis basically use time-varying Granger causality between an uncertainty index and MFA score based on Rossi and Wang [42] paper. This methodology is more powerful if the series have instabilities. In addition, classical Granger causality test does not allow to drive time-varying parameters. VAR-based reducedform Granger causality test requires stationarity in the data, and its test statistics is not valid if the series have potential structural breaks. Because of the nonstationary nature and existence of structural breaks in **Figure 1**, Rossi and Wang's [42] is the best methodology to analyze time-varying Granger causality with this data.

Following Rossi and Wang [42], Eq. (2) shows a reduced-form time-varying parameter VAR:

$$A\_t(L)\,\mathcal{y}\_t = \mathcal{u}\_t \tag{1}$$

$$A\_t(L) = I - A\_{1,t}L - A\_{2,t}L^2 - \dots - A\_{m,t}L^m \tag{2}$$

$$
\mu\_{t-}^{\
iid}(\mathbf{0}, \Sigma) \tag{3}
$$

where *Aj*,*t*,*j* = 1,…*m* show time-varying coefficients, [*yt*, *yt*−1,…, *yt*<sup>−</sup>*m*]′ show variables in the model, and *ut* shows error term.

The iteration of Eq. (2) and projection of *yt* onto the linear space created by (*y t*−1, *y t*−2,…, *y t*−*m*)′ derive the following equation:

$$\mathbf{y}\_t = \beta\_{1,t}\mathbf{y}\_{t-1} + \beta\_{2,t}\mathbf{y}\_{t-2} + \dots + \beta\_{m,t}\mathbf{y}\_{t-m} + \varepsilon\_t \tag{4}$$

where β*j*,*t*,*j* = 1,…*m* show time-varying coefficients, [*yt*, *yt*−1,…, *yt*<sup>−</sup>*m*]′ show variables in the model, and ε*t* shows heteroskedastic and serially correlated error term.

Finally, based on Eq. (4), robust Granger causality test figures out the validity of the null hypothesis in Eq. (5):

$$H\_0 \colon \mathfrak{h}\_t = \mathbf{0}\_t \\ for \ all \ t = 1, \ldots, T \tag{5}$$
